ABSTRACT Obstructive sleep apnea (OSA) is recognized as an extremely common disorder that results in excessive daytime sleepiness and an increased risk of automobile crashes. It is also an independent risk factor for insulin resistance, hypertension, cardiovascular events, increased cancer rates and cancer mortality, and accelerates the progression of neurodegenerative disorders. OSA is a systemic disorder and the cyclical intermittent hypoxia that occurs during sleep affects every organ system. OSA is heritable, with first degree relatives of individuals with OSA having a two-fold increased risk of the disorder. Although this has been known for over two decades, there are very limited data on gene variants conferring risk or protection for OSA. This is likely the result of underpowered studies and GENETIC HETEROGENEITY. There are multiple pathways to OSA. Obesity is a major risk factor, as are craniofacial dimensions. The relative risk of these different risk factors varies by ethnic group. There are also neuronal mechanisms and other physiological risk factors. This proposal seeks to directly address the issue of genetic heterogeneity using a very large sample of patients and state-of- the-art analysis techniques. Towards this end, we have assembled by far the largest sample of well characterized and genotyped patients with OSA. This is the result of collaboration with key sites in the eMERGE Network [Geisinger, Vanderbilt, Mayo Clinic, Northwestern], Kaiser Permanente Southern California, and the University of Pennsylvania. We propose to carry out a robust genome-wide association studies (GWAS) using an electronic health record (EHR) derived case definition and quantitative trait analysis of OSA severity. We will use novel machine learning approaches to specifically address genetic heterogeneity. The GWAS approach will be complimented by a PheWAS strategy as another approach to identifying common variants and related clinical phenotypes. Specifically, we will look for variants that associate with OSA and other clinical diagnoses in the EHR. It is likely that gene variants for OSA will have pleiotropic effects. Gene variants identified from GWAS and/or PheWAS will be further assessed. First, we will use available whole exome sequencing data to see if there are relevant rare variants in these genes, as well as to perform exploratory discovery analyses to look for novel rare variation. The functional role of the genes identified will be studied by functional network analysis. Finally, we will assess associations with quantitative data on intermediate traits identified by high throughput craniofacial and intra-oral photography. These data are currently available within the Sleep Apnea Global Interdisciplinary Consortium (SAGIC). Future directions will assess the function of the genes identified in model systems and the clinical utility of utilizing variants as part of risk stratification for OSA.